Back to skill
Skillv1.0.2
ClawScan security
Sentiment Priority Scorer · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignMay 1, 2026, 5:57 AM
- Verdict
- Benign
- Confidence
- high
- Model
- gpt-5.5
- Summary
- This instruction-only skill coherently scores provided real-estate leads and explicitly avoids writes, sends, reminders, or other state-changing actions.
- Guidance
- Before installing, confirm that you are comfortable sending lead text to your agent for scoring. The provided artifacts do not indicate writes, outbound messages, credential use, or hidden code execution.
Review Dimensions
- Purpose & Capability
- okThe stated purpose is limited to scoring and bucketizing normalized lead records, and the schemas match that scoring workflow.
- Instruction Scope
- okInstructions are bounded to validation, scoring, evidence generation, and output validation, with explicit prohibitions on writes, outbound sends, and action execution.
- Install Mechanism
- okNo install spec, binaries, environment variables, credentials, or code files are present; this is an instruction-only skill.
- Credentials
- okThe skill only asks to process user- or supervisor-provided lead data and does not request filesystem, network, database, or account access.
- Persistence & Privilege
- okArtifacts show no persistence, background behavior, privilege requirements, credential use, or mutation authority.